Here's the paradox defining AI in 2026: the more people use these tools, the less they trust them.
A new piece in The Atlantic argues that the first real AI crisis isn't technical, economic, or even political. It's psychological. And the data backs this up. ManpowerGroup's 2026 Global Talent Barometer, surveying nearly 14,000 workers across 19 countries, found that regular AI usage jumped 13% in 2025 while worker confidence in the technology dropped 18%. That's not a gap. That's a chasm.
The Numbers Paint a Clear Picture
Older workers are feeling this the hardest. Baby boomers reported a 35% decrease in confidence around AI, with Gen X close behind at 25%. But this isn't just generational technophobia. Fifty-six percent of workers globally said they received no recent skills development despite being expected to use AI tools daily. As Mara Stefan, VP of global insights at ManpowerGroup, put it: "Workers are being handed tools without training, context, or support."
Meanwhile, the business case isn't materializing the way the pitch decks promised. Only 10-12% of companies report actual revenue or cost benefits from AI implementation. A staggering 56% say they've gotten "nothing out of it." So workers are burning out (63% report fatigue from stress and heavy workloads), being asked to adopt tools they don't understand, and watching their employers fail to extract value from those same tools.
This Isn't About the Technology
The real problem is the deployment pattern. Companies rolled out AI tools the way they roll out everything: fast, with minimal training, and with a memo that amounts to "figure it out." That works fine for a new project management app. It does not work for tools that fundamentally change how people do their jobs and that occasionally produce confident-sounding nonsense.
The trust deficit has practical consequences. Workers who don't trust their AI tools either avoid using them (wasting the investment) or use them without verifying outputs (creating errors). Neither outcome is what anyone wants. And 64% of surveyed workers are "job hugging" - staying in roles despite burnout and dissatisfaction - partly because the AI transition has made the job market feel even more uncertain.
For anyone building workflows around AI tools, there's a lesson here. The bottleneck to AI productivity was never the model quality or the feature set. It's whether the person using the tool actually believes the output is reliable enough to act on. That's a training problem, a transparency problem, and increasingly, a management problem. The companies that figure out the human side of AI adoption will pull ahead. The ones that keep dropping tools on desks and hoping for the best will keep wondering why their AI investment isn't paying off.